Medical robotics has the potential to revolutionize how surgical procedures on bony anatomy are performed. It can assist surgeons by preparing bones much more accurately than mechanical guides or freehand cutting. It can help improve patient outcomes by decreasing surgical errors. In the orthopaedics community, though, medical robots have not been very successful due to a variety of issues ranging from robot size, to surgical time, and soft-tissue difficulties. We propose to overcome these difficulties by utilizing a miniature robotic milling device that attaches directly to the bone. Realizing the potential of minimally invasive procedures, the implant industry is currently in the process of redesigning implants and, together with surgeons, reexamining surgical procedures. It can be expected that the next generation of implants will be smaller and more suitable for eventual development of less invasive procedures. One example of such procedures in knee arthroplasty is patellofemoral resurfacing of the knee. Without lose of generality we will examine the capability of the robot in improving the accuracy of the femoral-component preparation for patellofemoral arthroscopy. From engineering point of this procedure simulates the future generation of orthopaedic arthroplasty where there will be a need to machine bone surface into a more complex shape that is not planar or spherical but complex surfaces. Therefore, the technology demonstrated by this research, though, will not be specific to the patellofemoral procedure, and will be adaptable to many other areas. We propose to develop a miniature robot that will be rigidly affixed directly to the bone. The robot itself will scan the shape of the femur directly, removing any need for preoperative imaging or intraoperative registration and tracking of the bone. With the additional input of the direction of patellar tracking, we will automatically optimize the planned position of the implant to ensure that it is properly aligned and congruent with the surrounding healthy cartilage. Congruency is a requirement for this procedure to make certain that the patellar component does not impinge on the edge of the femoral component and lead to early failure of the implant. The robot will then mill out the cavity to within 1mm of the planned location, guaranteeing complete coverage of the area with a defined surface uniformity. By validating the results generated with the proposed robot, first on wax blocks and then plastic bone phantoms, porcine bones, and finally on cadaver knees, we will show that the robot can deliver the accuracy required to precisely place the implant. This precise placement of the femoral component should reduce the possibility of impingement, patellar maltracking, and component loosening, and improve patient outcomes.